13
\$\begingroup\$

I need to measure the temperature in a terrarium and use that as an input to a temperature controller. The terrarium is quite humid, e.g. above 80%. Also the packaging should have minimal thermal mass, in order not to hinder the temperature controller, causing overshoots.

The sensor is a 1-wire IC

enter image description here

I have solved this problem by the means of hot silicone - first I solder the 3 leads of the chip to the wire, then I pour liquid silicon until no cavities are visible.

To my surprise, this does not work. The mean time between failures is 1 month. The microcontroller just begins reading a different, fixed temperature. I have another sensor, mounted in a dry box, that I use for sanity check. Therefore, I am sure that exactly the sensor fails.

How should I handle this in a DIY environment (i.e. not industrial)? Is there a more clever approach or just a correct technique to apply glue for waterproofing individual ICs?

\$\endgroup\$
2
  • \$\begingroup\$ Would switching to a thermocouple be possible? \$\endgroup\$
    – Matt Young
    Jan 13, 2014 at 13:11
  • \$\begingroup\$ @MattYoung, I like the digital thermometer, and have designed my firmware around it. Furthermore, I have not worked with thermocouples and designing the interfacing hardware would be very cumbersome for me. \$\endgroup\$
    – Vorac
    Jan 13, 2014 at 13:21

4 Answers 4

13
\$\begingroup\$

Depending on how much you value your time and effort, and reliability, here are two proposed options:

  1. Reliable and off-the-shelf: Purchase a DS18B20 sensor pre-fitted in a waterproof plastic or stainless steel case. The latter is currently on sale for $4.50 each:
    DS18B20 Waterproof

  2. Glass pill bottle and epoxy:

    • Solder the sensor leads to wires, and then seal the exposed metal of each lead with silicone heat shrink tubing.
    • Encapsulate the set of 3 wires in another larger silicone heat shrink tube.
    • To be extra-careful, wrap the entire assembly very tightly in several layers of plumber's Teflon tape: This tape is very thin and flexible, so it forms an excellent humidity seal - which is why plumbers use it for fittings!
    • Insert this bundle into the smallest glass pill bottle which it will fit.
    • Fill the pill bottle with waterproof epoxy (or any epoxy- or cyanoacrylate-based waterproofing sealant) to the top, thus submerging the bundle and some of the wire into this sealant
    • Follow the recommended curing procedure for the sealant used: The epoxy I use can be cured using UV light, so I expose it for an hour in a nail art UV lamp.
    • The last point of failure is the insulated wire itself, since typical wire used for electronics is not designed for sustained exposure to moisture. Use silicone rubber based, weatherproof insulated wire if possible. If not, use up the rest of the Teflon tape roll by wrapping it tightly around the wire bundle all the way out of the terrarium.
    • All done.
\$\endgroup\$
4
  • \$\begingroup\$ Maybe I shouldn't have purchased 75 units only because of some discount :D \$\endgroup\$
    – Vorac
    Jan 13, 2014 at 13:27
  • \$\begingroup\$ Oh, ouch! Maybe it's time to set up an eBay store and get rid of them! :-) \$\endgroup\$ Jan 13, 2014 at 13:28
  • \$\begingroup\$ Will that sealed sensor of the image (from a reputable supplier) really work underwater for long periods without additional sealing? I always wondered. \$\endgroup\$
    – FarO
    Aug 29, 2017 at 14:55
  • \$\begingroup\$ @OlafM The stainless steel one lasts. Not sure about the longevity of the plastic encased one, but that too is degassed epoxy potted, so it should do reasonably well. \$\endgroup\$ Aug 30, 2017 at 22:16
7
\$\begingroup\$

I have done much the same as this with a DS18B20 sensor, only I used it in a home made etching tank (heating element made with ceramic tiles and nichrome wire, old laptop supply and MOSFET with PWM from PIC16F690 using PID for control and simple display with LCD. It's been sitting in ferric chloride for over two years now and is still working fine.

Anyway, all I did was use a length of some clear PVC tubing (kind you get for a fishtank) and fed the sensor (with wires attached) to about an inch from the end. Then I added some clear silicone sealant to coat the sensor a bit for any expansion, and after that mixed some epoxy up and liberally filled the end of the tube with it until it had filled a fair bit past both sides of the sensor. I then added some heatshrink to the end of the tube and put a bit more epoxy in the small gap (this is probably overkill for the terrarium purposes though)

Anyway, it worked perfectly for me in a very hostile environment, so I would bet on it working for the terrarium.

A couple of pics to give a vague idea (sorry, best I could do as the end is glued to the bottom of the tank, and the tank has not been used regularly for a long time so the solution is old...)

enter image description here enter image description here

enter image description here enter image description here

\$\endgroup\$
5
  • \$\begingroup\$ +1 for simplicity - I'm way too accustomed to using glass pill bottles for deep water submersion, so this totally evaded me. \$\endgroup\$ Jan 13, 2014 at 14:46
  • \$\begingroup\$ My first try will probably be a length of fishtank hose (8mm diameter). Put the sensor inside and seal with silicone. Do I need epoxy, if I use enough silicon? \$\endgroup\$
    – Vorac
    Jan 13, 2014 at 14:57
  • \$\begingroup\$ I'd say yes, and heatshrink to cap the end too if possible (forgot to mention that, adding it now) If no epoxy then at least something similar. \$\endgroup\$
    – Oli Glaser
    Jan 13, 2014 at 15:10
  • \$\begingroup\$ @Anindo - it's just that I'd already done it that way and it worked nicely (I learnt about any humidity/moisture combined with a potential and also thermal expansion the hard way - amazing how quickly the legs on a TO-92 will disappear as soon as moisture gets to them :-) ) I think a glass pill bottle filled with epoxy seems very similar (as long as the wires are taken care of too) \$\endgroup\$
    – Oli Glaser
    Jan 13, 2014 at 15:25
  • \$\begingroup\$ @OliGlaser My application has a wide range of heat and cold, plus exposure to sunlight, and the pressure of up to 10 meters of water. PVC cracks in about 2 weeks under those conditions. However, for OP's requirement, your method is more than enough. \$\endgroup\$ Jan 13, 2014 at 15:28
2
\$\begingroup\$

Nothing is waterproof.
Some things are more waterproof than others :-).

Silicone rubbers are rather water permeable. What you are seeking to avoid is the presence of liquid water and air at the point where corrosion will occur. If you use a material with low water dissolved content and a tenacious voidless surface adhesion then you cannot get liquid water (no voids) and the dissolved water content is low so the reaction rate is low.

Many manufacturers make conformal coatings aimed at minimising water & oxygen related corrosion. Dow Corning is one such - better known than some and with a good reputation for quality products. I have no relationship with Dow Corning except as a customer.

Dow Corning Sylgard 184 is formulated specifically to meet the above requirements. It's main use if solar cell potting.

Dow Corning 1-2577 is a conformal coating which works reasonably well. It forms an ~= 0.1mm layer and can be sprayed brushed or dipped. Not breathing the fumes is a very good idea.

Dow Corming conformal coatings overviews:

Elastomeric

Electroplastic - 1.2577 is in this group

Solventless cure


Poormans CC: An initial layer of "polyurethane spray on clear coat varnish" may help. This sets by reacting with atmospheric water and is a low cost but reasonable conformal coating.

Clutching at straws: The presence of electrical potential in a corrosive environment will greatly accelerate corrosion. It may be that having the sensor powered is a significant contributor. If you are able to depower it for a significant portion of its cycle it MAY help.

EVA: I have not tried this but it has a moderately good chance of working. EVA is the traditional bonding and sealing agent of choice in Silicon on glass solar panels, which typically have lifetimes well in excess of 20 years. EVA goes a long way toward meeting the requirements mentioned above re water solubility, plus voidless adhesion to sealed surfaces. EVA is inserted as a plastic sheet between glass and silicon cells and then crosslinked under raised pressure and temperature. It seems likely [tm] that "just melting" EVA plastic over the sensor with a hot air gun has the prospect of meeting your needs. EVA plastic sheet is available from PV panel manufacturers or as high quality glasshouse (plastic house?) film . Laminating temperatures are usually in the mid 1xx range but low(er) temperature laminating EVA is available.

\$\endgroup\$
0
\$\begingroup\$

I think your method is fine, just the choice of material is wrong. Try an epoxy resin, apparently they have good heat conducting properties too. http://en.wikipedia.org/wiki/Epoxy#Electrical_systems_and_electronics

In one project we water proofed LED light panels with an epoxy resin and they seemed to hold up well. This was for lights on traffic signs though, not underwater. In another project we connected subconn marine connectors to some 4 core power cable, water proofing the join using 3M resin splices. These cables were deployed underwater for multiple 3 month periods (one during a cyclone) and the epoxy seal did not fail. Perhaps you could buy one of the smaller kits and embed the temperature sensor in there. Otherwise just get some epoxy and pot the sensor in a small ... pot.. like a pen cap.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.